Sexual violence, pregnancy and childbirth
Studies investigating the association of experienced sexual violence and outcomes in pregnancy and childbirth
Lena Henriksen
Women and Children’s Division Department of Obstetrics Oslo University Hospital – Ullevål
Faculty of Medicine University of Oslo
2015
© Lena Henriksen, 2015
Series of dissertations submitted to the Faculty of Medicine, University of Oslo No. 2001
ISBN 978-82-8333-010-6
All rights reserved. No part of this publication may be
reproduced or transmitted, in any form or by any means, without permission.
Cover: Hanne Baadsgaard Utigard.
Printed in Norway: AIT Oslo AS.
Produced in co-operation with Akademika Publishing.
The thesis is produced by Akademika Publishing merely in connection with the thesis defence. Kindly direct all inquiries regarding the thesis to the copyright holder or the unit which grants the doctorate.
Table of content
Summary... 4
Acknowledgements... 5
Definitions and Abbreviations ... 6
List of papers... 7
1. Introduction... 8
1.1 What is sexual violence? ... 8
1.2 Prevalence of sexual violence ... 9
1.3 Risk factors... 14
1.4 Sexual violence and health ... 14
1.5 Sexual violence and pregnancy-related health ... 14
1.6 Pathways ... 15
1.7 Pregnancy-related physiological symptoms ... 16
1.8 Hospitalisations during pregnancy ... 17
1.9 Mode of delivery and maternal outcomes ... 17
1.10 Neonatal outcomes ... 18
2. Study aims... 19
3. Materials and methods ... 20
3.1 The Norwegian Mother and Child Cohort study (MoBa)... 20
3.2 Exposure: Sexual violence ... 20
3.3 Overview papers I-IV... 21
3.4 Paper I – Aim, design, study population, variables and statistical analysis... 22
3.5 Paper II – Aim, design, study population, variables and statistical analysis . 26 3.6 Paper III– Aim, design, study population, variables and statistical analysis . 27 3.7 Paper IV– Aim, design, study population variables and statistical analysis.. 30
3.8 Missing... 31
3.9 Ethics... 32
4. Results... 33
4.1 Prevalence of sexual violence ... 33
4.2 Summary of results... 33
5. Discussion ... 35
5.1 Main findings ... 35
5.2 Consideration of methodology ... 35
5.2.1 Strengths and limitations of the studies ... 35
5.2.2 Selection bias... 36
5.2.3 Information bias... 37
5.2.4 Confounding ... 38
5.2.5 External validity ... 39
5.2.6 Causal inference ... 39
5.3 Interpretation of the results... 40
5.4 Clinical implications ... 43
5.5 Future research ... 43
6. Conclusions... 44
7. Errata... 44
8. References... 45
9. Appendix... 59
Summary
Background: Sexual violence against women is a recognised public health problem, and it is a phenomenon that persists in all countries regardless of value system and culture. There is growing evidence that women who have experienced sexual violence are at greater risk for negative health outcomes, even several years after the abuse. Pregnancy does not protect women from violence. For some women,
pregnancy may be the first time that they are aware of previous exposure to violence, or trauma that they have experienced may resurface.
Objectives: The first aim of our study was to explore whether a history of sexual violence was associated with pregnancy-related physical symptoms. Second, we investigated the relationship between lifetime sexual violence and antenatal hospitalisations. Third, we assessed the association between sexual violence and mode of delivery, and finally, we examined the relationship between a history of sexual violence and neonatal outcomes.
Subjects and methods: In this thesis, we used data from the Norwegian Mother and Child Cohort study (MoBa) linked to data from the Norwegian Medical Birth Registry. Studies I and II had a cross-sectional design and included 78 660 pregnant women. Studies III and IV were cohort studies that included 74 058 and 76 870 pregnant women, respectively. The participating women were recruited during their routine ultrasound examination, and they completed extensive questionnaires at 17 and 30 weeks gestation. History of sexual violence was reported at three levels of severity: 1) pressured into sexual acts (mild), 2) forced with violence (moderate), and 3) raped (severe). The comparison group was women not reporting sexual violence.
Results: In our studies, 12.0% of the women reported mild sexual violence, 2.8%
reported moderate sexual violence and 3.6% reported severe sexual violence (rape).
Compared with women who did not report a history of sexual violence, women who reported a history of sexual violence suffered from more pregnancy-related physical symptoms and were more often hospitalised during pregnancy. Severe sexual violence was associated with a higher risk of elective caesarean section (CS), and moderate sexual violence was associated with an increased risk of emergency CS.
There was no significant association between sexual violence and neonatal outcomes.
Acknowledgements
It has been a privilege to conduct this research and to be a PhD student. Many people have contributed to this thesis, and I would like to thank:
Mirjam Lukasse, my primary supervisor, for including me in this project. I have appreciated all the help and support I have received. I could not have done this project without you. I have enjoyed our discussions and your clear and direct supervision.
Berit Schei, my co-supervisor, for valuable comments and feedback and for sharing your great knowledge in this field.
Siri Vangen, project manager and co- author, for clear statistical and writing advice.
Britt-Ingjerd Nesheim, my co-supervisor, for valuable comments on the thesis.
Pernille Frese for much appreciated practical help.
Helse Sør-Øst for supporting this PhD with a three-year grant.
The Norwegian Institute of Public Health for giving me access to data from the Norwegian Mother and Child study.
The midwifery leaders at Oslo University Hospital, Kristi Hjelle, Hanne Knutsen, Sigrun Schei and former leader Grethe Næsje, for the support and encouragement to do a PhD.
My midwifery colleagues, study and travel friends, Birthe Ariansen and Elisabeth Grimsrud, for all the fun we have had while studying and travelling. You know you are the main reason that I started doing research at all.
All my colleagues at Føde/Barsel B Ullevål, for interest and encouragement when I first started this academic career.
My family and friends for your love and support and everything in life that is not about statistics and p-values!
Last, but not least, my gratitude goes to the Norwegian women who took the time to complete the comprehensive questionnaires in the Mother and Child Cohort study.
Definitions and Abbreviations
AOR Adjusted odds ratio
ASS Abuse Assessment Screen
BMI Body mass index
CI Confidence interval
CS Caesarean section
EU European Nation
EDA Epidural Analgesia
HPA Hypothalamic-Pituitary-Adrenal
LBW Low birth weight
IPV Intimate partner violence
MBRN Medical Birth Registry of Norway
MoBa The Norwegian Mother and Child Cohort Study
OR Odds ratio
PTB Preterm birth
PTSD Post-traumatic stress disorder
Q1 Questionnaire 1
Q3 Questionnaire 3
SCL-5 Hopkins Symptom Check List including 5 items SGA Small for gestational age
WHO World Health Organization
List of papers
This thesis builds on the following papers:
Paper I
Lukasse M, Henriksen L, Vangen S, Schei B. Sexual violence and pregnancy-related physical symptoms. BMC Pregnancy Childbirth 2012 11;12:83
Paper II
Henriksen L, Vangen S, Schei B, Lukasse M. Sexual violence and antenatal hospitalizations. Birth 2013 40:281-8
Paper III
Henriksen L, Schei B, Vangen S, Lukasse M. Sexual violence and mode of delivery:
a population-based cohort study BJOG 2014 121:1237-1244
Paper IV
Henriksen L, Schei B, Vangen S, Lukasse M. Sexual violence and neonatal outcomes: a Norwegian population-based cohort study. BMJ Open 2014;4:e005935 doi:10.1136/bmjopen-2014-005935
1. Introduction
Sexual violence against women is a significant and global public health problem.1-4 A recent report from the World Health Organization (WHO) states that 35% of women worldwide have experienced physical and/or sexual violence.1It is recognised that sexual violence has an adverse impact on women’s physical, psychological, behavioural and reproductive health,1,4-14including pregnancy complications.15-17Pregnancy does not protect women from violence, 18and the prevalence of physical or sexual violence during pregnancy ranges from 3-11% in high-income countries.19-21Several pathways are suggested between sexual violence and adverse health for pregnant women,1and women with a history of sexual violence can experience immediate and long-term morbidity.22,23A direct pathway of sexual violence can result in injury and immediate complications such as bleeding, rupture of membranes and preterm birth.1,24-26More indirect and complex pathways include physiological, psychological, behavioural and socio-economic factors.1,24,26, 27
Health-care providers need to understand the relationship between sexual violence and women’s ill health to be able to respond appropriately when they treat women with a history of sexual violence.1When women attend maternity care or other reproductive health services, an opportunity to support and help those who have been exposed may be present. In Norway, almost every pregnant woman attends antenatal care, a free and well-integrated part of the public health system,28and care encounters during pregnancy may represent an opportune time for investigation of sexual violence. During antenatal appointments, it may also be possible to give information about adequate interventions for women living in violent relationships.
Given the high prevalence of sexual violence, it is likely that a considerable proportion of pregnant women have been victims of sexual violence at some point during their lives. The majority of studies that have examined the association between a history of sexual violence and complications during pregnancy have primarily focused on childhood abuse or previous and current intimate partner violence, which can include physical and emotional abuse as well as sexual violence.17,29-36This thesis originated from a desire to better understand the consequences of lifetime sexual violence on pregnancy and childbirth. Though the literature is inconclusive, some studies have shown strong associations between adverse outcomes and sexual violence, including a 9-fold increase in the odds ratio of caesarean section among women raped as adults37and a greater than 3-fold increase in the odds ratio of premature delivery among women with a history of sexual violence.38Nevertheless, the majority of studies have shown smaller or no effects.1,29,30,36, 39-42
1.1 What is sexual violence?
To define sexual violence, it is necessary to first define violence. The WHO uses the following definition of violence:43
The intentional use of physical force or power, threatened or actual, against oneself, another person, or against a group or community, that either results in or has a high likelihood of resulting in injury, death, psychological harm, maldevelopment or deprivation (page 5.)43
This definition encompasses all types of violence and covers the wide range of acts that constitute violence, as well as outcomes beyond death and physical injuries.43 Violence affects men, women and children and is recognised as a public health problem and a violation of human rights.43
This thesis focuses on sexual violence, which is defined by the WHO as follows:
“ Any sexual act, attempt to obtain a sexual act, unwanted sexual comments or advances, or acts to traffic, or otherwise directed against a person’s sexuality using coercion, by any person regardless of their relationship to the victim, in any setting including but not limited to home and work” (page 149).43
Coercion can cover a range of acts such as psychological intimidation, blackmail or other threats of physical harm, and actual violence.43The definition also includes the inability to give consent due to intoxication, being drunk or being asleep.43Sexual violence includes rape, traditionally defined as vaginal, anal or oral sexual
intercourse obtained through force or threat of force.43-45Rape is usually defined as the most serious act of sexual violence.43Although both men and women are exposed to violence, women are more likely to be exposed to sexual violence, and perpetrators are usually men.46-48An important issue regarding sexual violence is the relationship between the victim and the perpetrator, and research has shown that a substantial proportion of violence occurs within intimate partner relationships such as marriage or cohabitation.3,49Sexual violence perpetrated by others, such as strangers, friends, teachers or colleagues, is usually referred to as non-partner sexual violence.2Intimate partner sexual violence and non-partner sexual violence are similar in terms of risk factors and health effects.50Nevertheless, there are some differences; sexual violence by an intimate partner may occur over a long time period, while rape by strangers may be a more violent single event.3,50,51
1.2 Prevalence of sexual violence
The investigation of sexual violence is a challenge from both an ethical and a methodological perspective.1,52Comparing studies is difficult, and prevalences vary due to both differences in settings and differences in the methods and measurements used.5,53Sexual violence is stigmatising; it may carry social sanctions for women who report it, and under-reporting is considered common.1,48This under-reporting makes it difficult to assess the actual prevalence. In addition, the focus has been on intimate partner violence, which typically includes several types of violence:
physical, emotional and sexual.2The development of a common definition and measurement tools for non-partner sexual violence has not received the same
attention as IPV.2When examining the prevalence of violence, it is common to use instruments that measure several types of violence.54For example, the Abuse Assessment Screen (AAS) measures physical, sexual and emotional abuse.55 Currently, stand-alone specialised surveys are considered the gold standard for obtaining valid data on sexual violence against women.1The 2000–2003 ”WHO multi-country study on women’s health and domestic violence against women”, one example of this type of survey, examined physical and sexual violence by an intimate partner at fifteen sites in ten countries,and the lifetime prevalence of intimate partner sexual violence ranged from 6% in city sites in Japan to 59% in Ethiopian provinces.56
A new worldwide prevalence of sexual violence against women can be found in the 2013 WHO report: ”Global and regional estimates of violence against women:
prevalence and health effects of intimate partner violence and non-partner sexual violence”.1This report states that 7.2% of women globally have reported non- partner sexual violence, and 30% have experienced physical and/or sexual violence by a partner.1In a Norwegian health survey that included all inhabitants in Oslo, 5.3% of the 8 643 participating women had been pressured into sexual acts as adults and 5% as children.57In another Norwegian study from 2008, Nerøien et al. found that 9.4% of the participating women had been exposed to sexual violence. This national study included 3 803 randomly selected women age 20–55 years.10Table 1 presents studies that report different prevalences of sexual violence.
Table 1. Population-based studies of lifetime sexual violence First author,
publication year and country
Characteristics of
the sample Method Sexual
violence Rape Thoresen
(2014)58 Norway
Population based, randomly sampled men and women N= 2 435 women
Telephone
survey 25.4% a 9.4%
Nationellt centrum för kvinnofrid, NCK (2014)59 Sweeden
Population based, randomly sampled men and women N= 5 681 women
Questionnaire based, paper and web
28% a 11% b
European union agency for fundamental rights, FRA (2014)60
All EU countries
Randomly sampled women from EU countries N=42 000
Face to face
interviews 11% 5%
Abrahams (2014)2 Worldwide
Estimates from 77
studies Systematic
review 7.2%c
11.5 %d NA MacDowall
(2013)61 Britain
Population based survey, men and women
Computer- assisted personal interviews
19.4% 9.8%
N=8869 women de Haas
(2012)62 Nederland
Population based survey, men and women
N= 3 283 women
Online
questionnaire 34% 12%
Steine (2012)47 Norway
Population based, randomly sampled men and women N=706 women
Questionnaire based
16.3% e NA
de Vries (2009)63 Nederland
Population based, randomly sampled men and women N=1 087 women
Telephone survey
NA 5.4%
Nerøien (2008)10 Norway
Population based, randomly sampled N=2 407 women
Questionnaire
based 9.4% f NA
Kilpatrick (2007)64 United States
A nationally representative sample
N=3 001 women
Computer- assisted telephone Interviewing
NA 18%
Garcia-Moreno (2006)56 WHO, Multi country
Randomly sampled women from 10 countries g N=24 097
Face to face
interviews 6 %-59% NA
Balvig (2006)65 Denmark
Population based, women
N= 3 552
Telephone
survey NA 9% h
Pape (2004)66 Norway
Randomly sampled men and women from Oslo N=2 355 women
Questionnaire
based 16% 5%
Tjalden (1998)67 United States
Population based, men and women N=8000 women
Telephone
survey NA 18% h
a Includes less severe violence
bIncludes severe sexual violence: forced intercourse or similar
c Worldwide prevalence non-partner sexual violence
dWestern European prevalence of non-partner sexual violence
eUnwanted sexual intercourse after age 16
fSexual violence in relationship
g Sexual IPV in fifteen sites in ten countries: Bangladesh, Brazil, Ethiopia, Japan, Namibia, Peru, Samoa, Serbia and Montenegro, Thailand, and the United Republic of Tanzania
hCompleted or attempted rape NA: Not Available
The reported prevalence of rape also varies, and studies report prevalences ranging from 3% to 20%.5,58,59,64,68An official Norwegian report from 2008 estimated that 8 000 to 16 000 women are exposed to rape or attempted rape in Norway every year.69 This inexact estimate, which was based on different prevalence studies at the time, reflects the difficulties and challenges inherent in measuring rape prevalence. The report stated that these estimates were conservative and that the majority of women exposed to rape or attempted rape do not report the abuse.69This finding is
supported by a 2014 Norwegian telephone survey that examined safety, violence and life quality in Norway.58In that study, one-third of the women reported that they had never told anyone about the abuse.58A national telephone survey from the US that examined women’s experiences with trauma and mental health reported similar results.70The prevalence of rape found in the recent Norwegian study was 9.4%
among the women who participated.58Table 1 shows how the rape prevalence varies in different studies.
Studies examining the prevalence of lifetime sexual violence in pregnant
populations report prevalences that range from 7% to greater than 30%.15,71-76In one Swedish study that examined sexual violence among women who attended antenatal care, 8.4% of the respondents were exposed to lifetime sexual violence.71In a cohort study from Denmark that included 2 638 low-risk nulliparous women, 9.2% had experienced lifetime sexual violence.73In a recent study that examined the prevalence of emotional, physical and sexual abuse among pregnant women in six European countries (Belgium, Iceland, Denmark, Estonia, Norway and Sweden), the prevalence varied from 8.3% to 21.1% for sexual abuse, with Iceland reporting the highest prevalence (21.1%) and Belgium reporting the lowest (8.3%).75The prevalence in Norway was 17.7%.75Two studies reporting prevalences greater than 30% are from the US, and both examined child sexual abuse only.15,76Table 2 shows the prevalences of lifetime sexual violence and sexual violence during pregnancy measured in pregnant populations.
Table 2. Prevalence of lifetime sexual violence in pregnant populations and sexual violence during pregnancy (studies on childhood abuse only are not included) First author,
publication year and country
Characteristics of the
sample Study design Sexual
violence Sexual violence during pregnancy Van Parys
(2014)77 Belgium
11 antenatal clinics N=1894
Cross- sectional
NA 1.4%a
Finnbogadóttir (2014)78 Sweden
Recruited at first antenatal visit N=1939
Cross-
sectional 15.7% 0.7%
Lukasse (2014)75 6 European countriesb
Unselected pregnant women
N= 7174
Prospective cohort
8.3-21.1% 0.4%
Groves Recruited at first Longitudinal NA 3.2%
(2014)79
South Africa antenatal visit
N=445 study
Salazar (2012)80 Nicaragua
All pregnant women in one municipal invited
N=478
Prospective
cohort 14% 7%
Schroll (2011)73 Denmark
Recruitment at antenatal clinics N= 2 638
Multi-centre
cohort study 9.2% NA
Nunes (2011)41 Brazil
Women attending antenatal care N=652
Prospective cohort
7.4% 0.5%
Silva (2011)81 Brazil
Pregnant women within Family Health Program
N=960
Prospective
cohort 5.7% 5.6%
Díaz- Olavarrieta (2007)82 Mexico
Women attending antenatal care N=1314
Prospective
cohort 10.1%c
3.7% d 1.8%
Van der Hulst (2006)72 Nedertand
Low risk pregnant women
N=625
Prospective
cohort 11.2% NA
Kaye (2006)32 Uganda
Women recruited in pregnancy at hospital N=612
Prospective
cohort NA 2.7%
Faramarzi (2005)38 Iran
Women at postpartum units N=3 275
Cross- sectional
NA 19.2%
Neggers (2004)42 US
Low-risk pregnant women
N=3 103
Prospective
cohort NA 2.4%
Johnson (2003)74 UK
Pregnant women at antenatal booking N=475
Cross-
sectional 10% NA
Stenson (2003)71 Sweden
Low risk pregnant women
N=1 038
Cohort 8.1% NA
Curry (1998)83 US
Abuse screening during pregnancy N=1 897
Prospective
cohort NA 4.5%
aSexual violence 12 months before pregnancy and/or in pregnancy
bBelgium, Iceland, Denmark, Estonia, Norway and Sweden
c Childhood
dLast 12 month NA: Not Available
1.3 Risk factors
Known risk factors for sexual violence are young age, drug and alcohol
consumption, having many sexual partners and having a former history of abuse.43,84 Violence in families often transmits across generations, and compared with non- exposed children, children who witness or experience violence are more likely to experience or to perpetrate violence as adults.54,85Low socioeconomic status is considered a risk factor for abuse 86as is single marital status.85In addition, societal gender power inequalities and hierarchical gender relations increase sexual violence, both violence that occurs within families, marriage and dating relationships and non- partner sexual violence.4,87
1.4 Sexual violence and health
Both physical and psychological health is affected by sexual violence.1Studies have shown that sexual violence is associated with the following: post-traumatic stress disorder (PTSD); anxiety disorders; depression; eating disorders; sleep disorders;
chronic pain such as headaches, abdominal pain, fibromyalgia and pelvic pain;
gastro-intestinal symptoms and sexually transmitted diseases.8,14,88-95Women with a history of sexual violence often seek help from health-care providers for various somatic symptoms, and they do not necessarily connect their problems to the history of violence.91,96-98It has been shown that risk-taking behaviour and self-destructive behaviour such as smoking and substance abuse are associated with the experience of sexual violence,26,99and women that have experienced sexual violence are more likely than women without a history of abuse to report poor quality of health.9,61,88,100
1.5 Sexual violence and pregnancy-related health
The general adverse health outcomes associated with sexual violence may also affect women in pregnancy. Pre-pregnancy health and negative health behaviours are likely to persist during pregnancy and affect health and health perception.71For some women, pregnancy may be the first time they are aware of previous exposure to violence, or trauma they have experienced may resurface.101,102The association between sexual violence and pregnancy-related health has been examined in terms of both maternal and neonatal outcomes.1,27,36,37,71,72,103,104The majority of studies have examined the effect of IPV 1,16,105,106or child sexual abuse.15, 29,34,35Studies indicate that a history of sexual violence is associated with more common complaints due to physical changes in pregnancy, vaginal bleeding, hyper-emesis, urinary tract infections and premature contractions.17,34,68Associations are also found between sexual violence and mode of delivery,35,76,103,107-109induction,103use of pain relief,103episiotomies35,72and anal sphincter tears.35In addition, studies have shown an association between sexual violence and preterm birth and low birth weight.1,29,38,110Nevertheless, the findings are inconclusive, and several studies have not found an association between sexual violence and complications during pregnancy and childbirth.15,30,32,40,71,72,104Most evidence regarding the association
between sexual violence and health for pregnant women comes from cross-sectional studies, which are unable to prove causality, or smaller studies without sufficient confounding control.1
1.6 Pathways
The possible pathways between sexual violence and adverse health outcomes are becoming better understood and documented.1Figure 1 shows the hypothesised pathways between sexual violence and pregnancy and birth complications based on the WHO’s conceptual framework regarding pathways and health effects from intimate partner violence.1The WHO suggests two main pathways, one direct and one indirect. The direct pathway involves injury to the woman that can lead to immediate complications such as miscarriage, ante partum haemorrhage, placental abruption, rupture of membranes and preterm birth.1,24,25,108,111The indirect pathway is mediated by stress and stress responses.112-114The literature provides evidence that the association between sexual violence and adverse health outcomes can be
explained by neural, neuro-endocrine and immune responses to chronic and acute stress.26,113,115It is shown that maternal exposure to stress can influence the hypothalamic-pituitary-adrenal (HPA) axis,114which may cause a shift in cortisone levels that can create an autoimmune/inflammatory response with effects such as chronic pain syndromes and inflammatory conditions.116Changes in these hormone levels may cause other negative outcomes such as premature delivery and foetal growth restriction.117,118
A psychological pathway is suggested that proposes that adverse outcomes can be mediated by mental distress.1,119,120This pathway may include conditions such as post-traumatic stress disorder (PTSD), somatisation, anxiety, fear of birth, an increased need for control and different birth strategies. Another indirect pathway is mediated by behavioural and other risk factors, and some women try to cope with the negative consequences of violence by using tobacco, alcohol or drugs,5,26,48all of which are risk factors for poor health.48Eating disorders are also reported among abused women.121Thus, some of the observed associations between sexual violence and adverse health may be related to these factors. Because most of the data on the health consequences of sexual violence are from cross-sectional studies, it is difficult to ensure the nature and direction of the relationship between sexual violence, the associated health-risk behaviour and the outcome.1,48Some of the possible pathways are based on the assumption that sexual violence may cause a particular behaviour or cause stress/mental distress that in turn increases the risk of adverse pregnancy outcomes.1,48
Figure 1. Model for possible pathways between lifetime sexual violence and adverse pregnancy outcome
Figure 1 is an adapted model based on the conceptual framework and pathways in the 2013 WHO report:”Global and regional estimates of violence against women:
prevalence and health effects of intimate partner violence and non-partner sexual violence”.1
1.7 Pregnancy-related physiological symptoms
Women can experience a variety of pregnancy-related symptoms such as nausea and vomiting, tiredness, backache, heartburn, constipation, vaginal discharge, leg cramps, oedema, headache, Braxton Hicks contractions, urinary incontinence, pelvic girdle relaxation, and urinary tract infections.122,123The majority of these symptoms are considered normal results of physiological changes caused by pregnancy.124-127 Usually, these symptoms have no effect on the outcome of the pregnancy, but they may cause discomfort and anxiety to women. Occasionally, these normal
discomforts may be symptoms of serious conditions.127Because of the association between sexual violence and general health as described in section 1.4, it is also likely that women who have experienced sexual violence report more of these complaints. Only two previous studies have investigated the association between
Sexual violence
Physical
trauma Psychological
trauma
-Injury -Rupture of membranes - Placental abruption -Premature contractions/
birth
Increase in stress hormones
- Smoking - Alcohol - High/low BMI - Inadequate antenatal care
Adverse outcomes in pregnancy and childbirth Mental
distress
-PTSD -Anxiety -Somatisation -Fear of birth
-Birth strategies:
control, flight, fight, surrender, retreat
sexual violence and pregnancy-related physical symptoms.30,34Both studies examined childhood sexual abuse and pregnancy-related complaints.
1.8 Hospitalisations during pregnancy
Even though pregnancy is generally considered a state of health rather than disease, pregnant women may experience severe complications due to both health risks before pregnancy and complications that occur during pregnancy.123There are several risk factors during pregnancy that require extra attention: hyperemesis;
bleeding; hypertensive disorders; premature contractions; sexually transmitted infection; and medical disorders such as asthma, epilepsy and diabetes mellitus.123 The prevalences vary between conditions and by when in pregnancy the conditions occur. For example, the prevalence of hyperemesis ranges from 0.8% to 3.2%,128,129 and approximately 8-12% of pregnant women are affected by hypertensive disorders.130,131Some complications may lead to hospitalisation, and because the general trend in pregnancy care is toward outpatient care,132a hospital admission suggests a complication of a certain degree of seriousness. In addition,
hospitalisation generates costs in terms of both health expenditures and an added burden for women and their families.19The prevalence of women being hospitalised during pregnancy for reasons other than childbirth varies; one population-based study from Canada, the results of which may be comparable to the situation in Norway, reports a prevalence of 5.7%.133The impact of sexual violence on antenatal hospitalisations has been poorly investigated. One study that investigated childhood sexual abuse found an increased risk for antenatal hospitalisations,29and there are studies that have found an association between IPV during pregnancy and antenatal hospitalisations.38,134, 135
An important issue is that the prevalence of sexual violence is actually much higher than the prevalence of other pregnancy complications in some settings.
Nevertheless, it receives considerably less attention within prenatal care.21,86
1.9 Mode of delivery and maternal outcomes
The proportion of childbearing women that is defined as being high risk is increasing,136and technological developments are allowing pregnancies in women with conditions that previously prevented conception or successful pregnancies.136 How clinicians address these complications differs, but there is global concern about the overuse of interventions.137Although advanced maternal age, multiple gestation and other medical factors can increase pregnancy complications, they do not fully explain the increased rate of interventions.138Unnecessary interventions during pregnancy and birth seem to be increased in high-income countries, which may cause further complications for women and newborns.139In addition, the economical costs of increased interventions are substantial.140
The increased caesarean section rate is the main cause for concern.141Rates greater than 10-15% are considered not beneficial to the health of the mother or child.142,143 The caesarean section rate exceeds this recommendation in all countries in the world with the exception of sub-Saharan Africa.141While a CS may save the life of both mother and child, the surgery is not without the risk of short- and long-term complications.144Immediate complications include infections or thrombosis,144and long-term complications include increased risk of preterm birth, stillbirth, placenta accreta, bleeding and rupture of the uterus due to the operation performed in earlier pregnancies.145
In Norway, the CS rate was 16.8% in 2012 with regional differences ranging from 12.8% to 21.7%.146In addition to the increase in risk factors among pregnant women in general,136, 138improvements in anaesthetics and operative techniques and a lower threshold among obstetricians for performing CS147may have contributed to the increase in CS rates. Another suggested explanation is the change in how women participate in medical decision-making and their preferences for delivery.147,148 Some literature has suggested that the increase in maternal requests for CS may be partially due to fear of childbirth.147,149An association between sexual violence and fear of childbirth has been reported.36,150Women with a history of sexual violence may thus wish to have a CS because of their past negative experience.
Induction of labour is also an increasingly used intervention.151In Norway, approximately 18% of all births are induced.152In the case of both an elective CS and induction of labour, the start of birth is planned, which may be a way for women to have more control of the birth process. Control is one suggested coping
mechanism for abused women during childbirth.153The bodily experience of childbirth may trigger memories of sexual abuse and affect a woman’s ability to cooperate with staff in the second stage and thus may be associated with vaginal operative deliveries and perineal trauma, such as episiotomies and anal sphincter tears.154
1.10 Neonatal outcomes
The neonatal outcomes investigated in this thesis are preterm birth, low birth weight and small for gestational age. Preterm birth is a common health problem,155,156and prematurity is considered the leading cause of death for newborns.156Approximately one in ten babies are born preterm worldwide.156Low birth weight (LBW) can be a consequence of preterm birth (PTB) or intra uterine growth restriction, the latter leading to the birth of small for gestational age (SGA) infants.157There are some biological risk factors for PTB and LBW: multiple pregnancies, a previous preterm birth and uterine or placental abnormalities.156,157Other important but less
understood factors for PTB and LBW are behavioural and social factors,156
including maternal age, socio-economic status, ethnicity, maternal weight, substance abuse, stress, depression and violence.117,155,156The prevalence of preterm birth was 5.5% in Norway in 2011,152a prevalence that has remained low and stable for the last decade.158
2. Study aims
The overall aim of this study was to examine the effect of lifetime sexual violence on women’s health during pregnancy and delivery. We wanted to examine whether a history of sexual violence was associated with adverse maternal and neonatal outcomes.
The following research questions were defined:
Paper:
x What is the prevalence of sexual violence in the Norwegian Mother and Child cohort?
I
x Are women with a history of sexual violence more affected by
pregnancy-related physical symptoms than women without a history of sexual violence?
x Are women who have been exposed to lifetime sexual violence hospitalised more frequently during pregnancy than non-exposed women?
II
x Are there differences between exposed and non-exposed women regarding reasons for hospitalisations?
x Is there an association between lifetime sexual violence and mode of delivery?
III x Do women with a history of sexual violence have more adverse maternal outcomes than non-exposed women?
x Are there differences in birth weight and gestational age among women exposed to sexual violence compared with women without such a history?
IV x Do women who have been exposed to sexual violence have an
increased risk of giving birth prematurely or giving birth to a child with low birth weight or a small for gestational age child?
3. Materials and methods
3.1 The Norwegian Mother and Child Cohort study (MoBa)
The Norwegian Mother and Child Cohort Study is a prospective population-based cohort study conducted by the Norwegian Institute of Public Health between 1999- 2008.159The study was conceptualised in the 1990s, and the main aim of the study was to find causes of disease.159Pregnant women were recruited at the time of their routine ultrasound examination (approximately 17 weeks gestation). All pregnant women in Norway were eligible to participate in the study, and 40.6% of the invited women consented to participate. The MoBa leader group has upgraded the
participation rate during the time we conducted these studies. Therefore, we had a participation rate of 38.7% in Papers I-III. The fathers of the children were also invited to participate. Data were collected from several questionnaires both in pregnancy and postpartum and from biological material. The cohort includes approximately 109 000 children, 91 000 women and 71 700 men. Additional information about the MoBa study can be found at the following web address:
http://www.fhi.no/studier/den-norske-mor-og-barn-undersokelsen.
The participants in this study received a postal invitation at their routine ultrasound appointment. During pregnancy, the women answered questionnaires focused on demographic factors, general health, reproductive history and maternal health during pregnancy. We used questionnaire 1 (Q1), completed at approximately gestational week 17, and questionnaire 3 (Q3), completed at approximately gestational week 30.
Data from the MoBa study were linked with data from the Medical Birth Registry of Norway (MBRN), a registry that maintains record of all deliveries in Norway based on a standardised form completed by midwives shortly after delivery.160
This study is based on version VI of the quality-assured data files released for research in 2011.
3.2 Exposure: Sexual violence
The exposure variable was collected from Q1 and was used in all four papers included in this thesis. The women were asked if they had ever been pressured or forced into sexual relations. The answer options included the following: 1) No, never 2) Yes, pressured 3) Yes, forced with violence 4) Yes, raped. A positive answer was defined as having experienced sexual violence. Women with more than one positive answer were classified according to the most severe level of violence reported. The answer options were then recoded into mild, moderate and severe sexual violence. We used this terminology because it corresponds to other studies that have used validated instruments to study the prevalence of violence.56We recognise that all three answer options may be considered severe by the person who experienced the violence. A study unit in MoBa is a pregnancy, and a study unit in our study is a woman; therefore, we excluded pregnancies in women who
participated more than once. In so doing, we ensured that the exposure was counted only once for each woman.
Women could also indicate the timing of the violence at the following time periods:
1) during this pregnancy, 2) during the last six months before pregnancy or 3) earlier than six months before pregnancy. The 1 712 women who responded to the first version of Q1 had the option to answer ‘earlier’ and ‘during the last 12 months’. We therefore created the variables ‘previous’ and ‘recent sexual violence’, with the latter including sexual violence that occurred during last 12 months including the current pregnancy. Table 1 shows the questions on sexual violence in the MoBa study.
Table 3. Questions on sexual violence in The Norwegian Mother and Child Cohort Study (MoBa)
Version A, answered by 1 712 women:
Have you ever been pressured or forced into sexual relations?
(Fill in one or several boxes.)
Last 12 months Earlier
No, never....………...
Yes, pressured………
Yes, forced with violence….….
Yes, raped……….….
The questions used in the other versions:
Have you ever been pressured or forced into sexual relations?
(Fill in one or several boxes.) During this
pregnancy
Last 6 months before pregnancy
Earlier No, never....………
Yes, pressured………
Yes, forced with violence….…..
Yes, raped……….….
All the questionnaires used in the MoBa study are available at the following web address: http://www.fhi.no/studier/den-norske-mor-og-barn-
undersokelsen/sporreskjemaer.
3.3 Overview papers I-IV
Table 4 provides an overview of the sample, exposure, main outcome variable, covariates, design and statistical analysis in papers I-IV; the outcome measures and covariates are further described in sections 3.4 to 3.7.
Table 4. Overview papers I-IV
Paper I Paper II Paper III Paper IV
Sample 78 660 78 660 74 058 76 870
Exposure Sexual violence Sexual violence Sexual violence Sexual violence Outcome
variables
Pregnancy-related physical
symptoms
Antenatal
hospitalisations Mode of delivery
Maternal outcome Gestational age, Birth weight Covariates Age, parity,
education, occupation, civil status, smoking, alcohol, BMI, mental distress, child and adult physical abuse, child and adult emotional abuse, child and adult sexual abuse
Age, parity, education, occupation, civil status, smoking, alcohol, BMI, mental distress, child and adult physical abuse, child and adult emotional abuse
Age, parity, education, occupation, civil status, smoking, BMI, mental distress, child and adult physical abuse, child and adult emotional abuse, maternal diabetes, pre- eclampsia, macrosomia, previous CS, induction of labour, dystocia, epidural
Age, parity, education, smoking, BMI, mental distress, child and adult physical abuse, child and adult emotional abuse, spontaneous or provider-initiated start of labour
Design Cross-sectional Cross-sectional Prospective cohort Prospective cohort Statistical
analysis Descriptive, Binary logistic regression
Descriptive, Binary logistic regression
Descriptive, Binary and multinomial logistic regression
Descriptive, Linear regression, Binary logistic regression
3.4 Paper I – Aim, design, study population, variables and statistical analysis 3.4.1 Aim
The aim of Paper I was to investigate whether a history of sexual violence was associated with pregnancy-related symptoms. In addition, we wanted to explore whether women with a history of sexual violence suffered longer or to a greater extent from the reported symptoms compared with women without such a history.
3.4.2 Design
Although MoBa is a cohort study, Paper I has a cross-sectional design. In this study, both the exposure and outcome data were collected at the same time.
3.4.3 Study population
Data were available on 92 838 pregnancies from Q1, Q3 and MBRN. Women were the observation unit in our study; we therefore excluded 13 475 pregnancies in women who participated more than once. We excluded 703 women who had not answered the question on sexual violence in Q1, yielding a study sample of 78 660 women
Figure 2. Flow chart for study I.
3.4.4 Outcome variables
Pregnancy-related physical symptoms: All symptoms were derived from MoBa, Q1 and Q3, except for Braxton Hicks contractions and leg cramps, which were only reported in Q3. Women indicated if they were “not at all”, “a little bothered” or
“bothered a lot” by Braxton Hicks contractions. For the remaining physical symptoms—backache, tiredness, constipation, pelvic girdle relaxation, heartburn, nausea and vomiting, oedema, candidiasis, pruritus gravidarum, leukorrhoea, headache, urinary tract infection and urinary incontinence—women reported the number of 4-week periods during which they were bothered by each symptom. The number of 4-week periods women could choose varied from 8 for most symptoms (total of 32 weeks starting from 0–4 weeks of pregnancy) to 5 for leg cramps (only asked about in Q3, which started at 13–16 weeks of pregnancy).
3.4.5 Covariates
Age: Maternal age, defined as age in years at delivery, was taken from Q1. Age was considered a true confounding factor and included in all papers.161In Paper I,
92 838 pregnancies for which data were available from Q1, Q3 and MBRN
13 475 pregnancies in women who participated twice or more
79 363 women with data from Q 1,3 and MBRN who participated once
703 women who did not answer any of the questions on sexual violence in Q1 78 660 women in the
study sample
age was categorised into 5 groups: younger than 19 years, 20–25 years, 26–31 years, 32–37 years or 38 years and older.
Parity: Parity was taken from Q1 and defined as previous births after 21 completed weeks of gestation. Parity was dichotomised into nulli- and multiparous women. We have examined the effect of parity in all studies because parity is considered to be associated with the outcomes of interest in this thesis.162
We used education and/or occupation as a proxy for socio-economical status. Low socio-economical status is considered a consistent predictor of both violence and pregnancy outcomes;85,163therefore, we examined these factors in all the papers.
Education: Education was taken from Q1, and we used education in years FDWHJRULVHGLQWRJURXSVSULPDU\\HDUVVHFRQGDU\\HDUV\HDUV beyond secondary (13–16 years anGDQG!\HDUVEH\RQGVHFRQGDU\\HDUV Occupation: In Q1, the women were given 11 possible choices for occupation: 1) student, 2) at home, 3) intern/apprentice, 4) military service, 5) unemployed/laid off, 6) rehabilitation/disabled, 7) employed in public sector, 8) employed in private sector, 9) self-employed, 10) family member without steady income in family company (e.g., farming, business) and 11) other. Three categories were made:
student (including answer options 1 and 3), employed (including answer options 4, 7, 8, 9, and 10) and unemployed (including answer options 2, 5, and 6).
Civil status: The women were asked about their civil status in Q1, and the answers were coded as either married/cohabitant or single, which also included both divorced and widowed. We included civil status because research shows that being single is associated with an increased risk of being exposed to violence.85
Pre-pregnancy body mass index (BMI):We controlled for BMI in all studies because both low and high BMI are considered to have adverse effects on pregnancy and childbirth,164and BMI is associated with the exposure.121,165BMI, derived from Q1, was the pre-pregnancy BMI grouped into 4 categories: <20, 20-24.9, 25.0-29.9 RUNJP
Behavioural factors such as smoking and alcohol consumption are well-known factors associated with a history of violence.1,166,167They are also known risk factors for several pregnancy-related physical symptoms and complications.123,124 Smoking: Smoking was derived from Q1 and categorised as either smoking during pregnancy or not. Smoking during pregnancy included both daily and occasional smoking.
Alcohol: Alcohol consumption was taken from Q1 and dichotomised into any or no alcohol use during pregnancy.
Mental distress:Mental distress is associated with both sexual violence1,119,120and pregnancy complications117,168and was therefore included in all papers. The
Hopkins Symptom Checklist (SCL-5), which accounted for five items from Q3, was used to define symptoms of mental distress with a cut-off of t2.0 points, as suggested by Strand.169
Other type of abuse:Because of the co-occurrence of different types of violence,56 we examined the effect of physical and emotional abuse both as a child and as an adult. Information on adult physical violence was taken from Q1 and consisted of a positive answer to whether women had experienced being slapped, hit, kicked or otherwise bothered in a physical manner as an adult. Child physical violence was taken from Q3 and consisted of a positive answer to the question: “Did you experience physical violence before the age of 18?” Emotional abuse as a child RUDVDQDGXOWFRQVLVted of a positive answer to either or both of the following questions in Q3: “Has someone over a long period of time systematically tried to subdue, degrade or humiliate you?” or “Has someone threatened to hurt you or someone close to you?” Child sexual abuse consisted of a positive answer to a question in Q3 in which women were asked if they had been pressured into sexual acts/activities as a child (<18 years). In addition, women were asked in Q3 if they had been pressured into sexual acts/activities as anDGXOW\HDUV7KHVH[XDO violence reported in Q1 could be the same act as the violence reported in Q3.
Because the question on sexual violence is more detailed in Q1 and because its wording is less likely to include non-contact sexual abuse, we selected our exposure variable from Q1. Testing for collinearity between sexual violence reported in Q1 and sexual abuse reported in Q3 resulted in a Pearson’s correlation coefficient of 0.605, which is above the generally accepted cut-off of 0.4 for including the variable as a covariate in regression analyses.170Due to this overlap and collinearity, sexual abuse from Q3 was not entered in the regression models.
3.4.6 Statistical analysis
Frequency analyses were used to quantify the prevalence of each level of sexual violence. Cross-tabulations and Pearson’s chi-square tests were used to study percentages and to assess differences in demographic and other characteristics for women reporting sexual violence compared with women not reporting sexual violence. Binary logistic regression analyses were conducted to estimate the crude and adjusted ORs, the 95% CIs for the association between the different levels of sexual violence, and the 90thpercentile of the number of 4-week periods of suffering for each pregnancy-related physical symptom. To estimate the
independent associations between sexual violence and the reporting of pregnancy- related physical symptoms, we adjusted for the other types of violence and abuse reported, as well as age in all adjusted models provided there were enough cases (model 1). We made two additional models. In model 2, we adjusted for mental distress. In model 3, we added the a priori selected covariates: pre-pregnancy BMI, parity, smoking and alcohol consumption in early pregnancy. We also examined the association between the timing of the abuse and suffering from 8 or more pregnancy-related symptoms. The comparison group for all analyses was women not reporting sexual violence in all four studies. All analyses were two-VLGHGDWĮ 0.05 and conducted with the statistical program SPSS version 18.0 or 19.0.
3.5 Paper II – Aim, design, study population, variables and statistical analysis 3.5.1 Aim
We studied whether a history of sexual violence was associated with an increase in antenatal hospitalisations. We also explored the reasons for hospital admissions.
3.5.2 Design
Paper II has a cross-sectional design.
3.5.3 Study population
Figure 2 in section 3.4.3 shows the study population. The total sample comprised 78 660 women.
3.5.4 Outcome variables
Antenatal hospitalisation during pregnancy: Information on antenatal
hospitalisations was derived from two composite questions in Q3. The women answered an overall question about whether they had been hospitalised during pregnancy. As part of this question, they were asked to select one or more reasons for hospitalisation from the following options: prolonged nausea and vomiting (hyperemesis), bleeding, leaking of amniotic fluid, threat of preterm birth, high blood pressure, pre-eclampsia and other. Due to overlap, high blood pressure and pre-eclampsia were recoded into one variable called hypertensive disorders. The 1 063 women who answered “yes” to the overall question about hospitalisation without giving any specific reason were classified as “Admitted without reported reason”.
In addition, women could indicate the time periods during which they were hospitalised; they selected 4-week periods from 0–4 to 29+ weeks gestation. The variable “Admitted more than once for different reasons” included women who reported admission in two time periods for different reasons. “Admitted more than once for the same reason” consisted of women who reported admission in two or more time periods for same reason.
3.5.5 Covariates
The following covariates were included in this study: parity, education,
occupation, civil status, use of alcohol, smoking during pregnancy, BMI, mental distress and other types of abuse as described in section 3.4.5.
3.5.6. Statistical analysis
Frequency analysis was used to quantify the proportion of the different levels of sexual violence and the prevalence of the different outcomes. Cross-tabulations and Pearson’s chi-square tests were used to assess differences in characteristics between
women who were and were not hospitalised. Fisher’s exact tests were used when the assumptions for the chi-square test were not met. The associations between exposure variables and the outcomes were estimated as crude and adjusted ORs using binary logistic regression analyses with 95% CIs.
In the preliminary analysis, we controlled for potential confounding factors and other covariates: socio-demographic characteristics (age, parity, education, occupation and civil status); behavioural factors (smoking and alcohol), BMI and other types of abuse. These factors are all considered to be associated with a history of violence and are also know risk factors for pregnancy-related complications that can lead to hospitalisation.99,171-173Age, parity, BMI, mental distress and other types of abuse were kept in the final adjusted models. The other covariates did not influence the OR.
3.6 Paper III– Aim, design, study population, variables and statistical analysis 3.6.1 Aim
The primary aim was to investigate whether a history of sexual violence was associated with the mode of delivery. We also examined the association between sexual violence and selected maternal outcomes.
3.6.2 Design
Paper III was a population-based cohort study in which women were followed from approximately gestational week 17 until they gave birth.
3.6.3 Study population
Only full term births were included in this study. Figure 1 describes the inclusion and exclusion criteria. The study sample comprised 74 058 women.
Figure 3. Flow-chart for study III:
3.6.4 Outcome variables
Mode of delivery: All outcome variables were obtained from the MBRN. The mode of delivery was classified as spontaneous birth, instrument-assisted vaginal delivery (vacuum- or forceps-assisted births), elective caesarean section (CS) and emergency CS. Elective CS included CSs that were planned >8 hours prior to delivery. Emergency CS included all other caesarean deliveries.
Other maternal outcomes: Induction, epidural, dystocia, episiotomy and anal sphincter tears were also studied. Dystocia was recoded from a variable in MBRN that consisted of the following: dystocia, foetopelvic disproportion, abnormal labour and augmentation.
92 838 pregnancies for which Q1, Q3 and MBRN data were available
13 475 pregnancies in women who participated twice or more
79 363 women with Q1, Q3 and MBRN data who participated once
703 women who did not answer any of the questions on sexual violence in Q1 78 660 women with Q1,
Q3, MBRN who answered question on sexual violence
74 058 women in the study sample
4296 with premature birth
306 with no information on gestational length
3.6.5 Covariates
Age:We controlled for younger (\HDUVDQGROGHUPDWHUQDODJHEHFDXVH both are associated with mode of delivery and other birth outcomes.151,174
Parity: Parity was obtained from the MBRN and coded as nulli- and multiparous women. In this study, we stratified for parity because nulli- and multiparous women are considered to be different in terms of both mode of delivery and the other outcomes. The majority of other studies that have addressed the same outcomes have examined nulliparous women only, and by stratifying, we were able to compare our study to the others.
The following variables were also included: education, occupation, civil status, use of alcohol, smoking during pregnancy, BMI, mental distress and other types of abuse.
Other risk factors: We included risk factors that were considered to be associated with mode of delivery and the different maternal outcomes.138,151Information about these variables was obtained from the MBRN and included pre-eclampsia, maternal diabetes (all types), macrosomia (birth weight over 4.5 kg) and previous CS for multiparous women. Induction, dystocia and epidural were considered covariates associated with the mode of delivery when they were not the outcome of interest.
3.6.6 Statistical analysis
Cross-tabulations, Pearson’s chi-square tests and linear-by-linear associations were used to calculate percentages and to assess differences in demographic and obstetric factors for women with a history of mild, moderate and severe sexual violence.
Multinomial logistic regression analysis was used to examine the association between sexual violence and mode of delivery. Univariable models with the mode of delivery as the dependent variable and mild, moderate and severe sexual violence as the independent variables were performed first. The adjusting variables were then added in sequence to the preliminary univariable models. Variables were included in a multivariable model if they were associated with either of the outcomes with a p- value of 0.1 or less. Variables that retained a significant association with either of the outcomes in this step were included in the final models. A p-value <0.05 was the level of inclusion for the adjusting variables in the final multivariable models.
Binary logistic models were similarly performed, with the dependent variables comprising binary variables.
3.7 Paper IV– Aim, design, study population variables and statistical analysis 3.7.1 Aim
In this study, we assessed the relationship between sexual violence and both gestational age at birth and birth weight. Additionally, we explored the associations between sexual violence and PTB, LBW and SGA.
3.7.2 Design
Paper IV was a population-based cohort study.
3.7.3 Study population
Figure 4 shows the inclusion and exclusion process for this study. The study sample comprised 76 870 women.
Figure 4. Flow-chart for study IV Pregnancies for which Q1, Q3 and MBRN data were available N=92 838
13 475 pregnancies of women who participated twice or more
Women with Q1, Q3 and MBRN data who participated once N=79
Women who answered questions on sexual violence N=78 660
76 870 women in the study sample
703 women who did not answer any of the questions on sexual abuse in Q1
Multiple births: N=1389
Missing gestational duration: N= 297 Gestation <22 weeks N=7
Gestation >44 weeks N=46 Missing birth weight N=41 Birth weight <500 g N=6 Birth weight >6000 g N=4
3.7.4 Outcome variables
Neonatal outcomes: Neonatal outcomes were obtained from the MBRN. Gestational age at birth in days was based on ultrasound at approximately gestational week 18.
For women with no ultrasound (1.7%), the gestational age was based on the last menstrual period. PTB was defined as gestational age <37 weeks, LBW as a birth weight <2500 grams, and SGA as birth weight below the 10th percentile for gestational age at birth. SGA was calculated using the Norwegian-specific foetal growth tables by Skjerven et al.175
3.7.5 Covariates
We used the following socio-demographic and behavioural variables: age, parity, education, smoking and BMI. In this study, age was categorised into 5 groups:
younger than 20 years, 20–24 years, 25–29 years, 30–34 years or 35 years and older. In addition, we controlled for mental distress and other type of violence and abuse.
3.7.6 Statistical analysis
Frequency analysis and cross-tabulation were used to assess characteristics that were presented as percentages within the entire sample and the different outcomes. Linear regression was conducted to assess differences in birth weight and gestational age for children born to women with and without a history of mild, moderate and severe sexual violence. The association between sexual violence and PTB, LBW and SGA was estimated with crude and adjusted OR using binary logistic regression analyses.
All analyses were adjusted for maternal age, parity, education, smoking, BMI and mental distress in the first step. Birth weight was additionally adjusted for gestational age. We further adjusted for other types of violence in the second step.
We stratified the sample into spontaneous start of birth and provider-initiated start of birth (induced start of birth or elective caesarean section) for gestational age because a provider-initiated start could influence the gestational age at birth. Information on the initiation of delivery was taken from the MBRN.
3.8 Missing
In all studies, the prevalence of missing data was less than 4% for all variables except alcohol consumption, for which the prevalence of missing data was
approximately 12%. The missing data for alcohol consumption were recoded into a dummy variable and included in the regression as a categorical variable to prevent the exclusion of a large number of women from the analysis. Because missing data was not a significant problem, no imputing methods for missing data were used176 except for the missing data for the SCL-5 (approximately 3%), which were replaced by the series mean. The results of the logistic regression analyses remained
approximately the same when performed with the complete exclusion of missing data compared with using the imputed missing data for SCL-5.
3.9 Ethics
The MoBa study was approved by the Regional Committee for Medical Research Ethics and the Norwegian Data Inspectorate. All participants were informed about the purpose of the study and that their participation in the study was voluntary. Each participant signed a written consent approving the use of the data provided for research purposes and linking of the data with the MBRN. The participants were free to withdraw from the study at any time. All personal identifiers were removed in the main database, and no personal data were sent to the researchers.
Some of the participants may have found some questions intrusive, and the inclusion of questions on sexual violence was a controversial issue in the MoBa study. The participants were asked to complete the questionnaires as fully as possible, but they were not contacted if any of the questions were incomplete. The ethical
considerations regarding MoBa were addressed in special meetings prior to the start of the study. One concern was the safety of the women that may live with a violent partner. The safety of the respondents is paramount when examining violence.177 Therefore, questions that included the partner as the perpetrator were excluded to reduce the risk for women filling out the questionnaires if they were living with an abusive partner. The recruiting hospital provided contact information to women upon request.
No intervention was done in the MoBa study, and this reduced potential disadvantages. Questions about sensitive topics, such as sexual violence, may contribute to negative feelings including self-blame, stigmatisation or humiliation.178 Nevertheless, studies show that women are willing to answer questions about abuse and they report meaningfulness about their participation in studies with questions about sensitive topics.179
